Grasp as You Say: Language-guided Dexterous Grasp Generation

Authors: Yi-Lin Wei, Jian-Jian Jiang, Chengyi Xing, Xian-Tuo Tan, Xiao-Ming Wu, Hao Li, Mark Cutkosky, Wei-Shi Zheng

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments are conducted on Dex GYSNet and real world environments for validation.
Researcher Affiliation Academia 1 School of Computer Science and Engineering, Sun Yat-sen University, China 2 Stanford University, USA 3 Key Laboratory of Machine Intelligence and Advanced Computing, Ministry of Education, China
Pseudocode No The paper describes methods and frameworks in text and diagrams but does not include explicitly labeled 'Pseudocode' or 'Algorithm' blocks.
Open Source Code No We promise to release all code and the complete dataset after the publication of this paper.
Open Datasets Yes We first collect object meshes and human grasps data from existing datasets [27].
Dataset Splits Yes We split the Dex DYSNet dataset at the object instance level, using 80% of the objects within each category for training and 20% for evaluation.
Hardware Specification Yes All experiment are implemented with Py Torch on a single RTX 4090 GPU.
Software Dependencies No The paper mentions 'Py Torch' as the implementation framework but does not provide specific version numbers for it or any other software dependencies.
Experiment Setup Yes For training our framework, the training epochs are set to 100 for intention and diversity grasp component and 20 for Quality Grasp Component. The loss weights are configured as follows: λ2 para = λ3 para = 10, λ2 chamfer = λ3 chamfer = 1, λ3 cmap = 10, λ3 pen = 100, λ3 spen = 10. Throughout all training processes, the model is optimized with a batch size of 64 using the Adam optimizer, with a weight decay rate of 5.0 10 6. The initial learning rate is 2.0 10 4 and decay to 2.0 10 5 using a cosine learning rate [52] scheduler.